PhD Work: Design
Contents
- A Domain Ontology for a NLQS
- Dataflow for a NLQS
- Visualizing Ontologies
- Ontology and RDF Data Files Map
- NLP Systems
- An Ontology for Representing User-Controlled Thesauruses
- Datastores
Entries
<< Previous Entry | Next Entry >>
5. NLP Systems
As remarked upon in my Nine Month Progress Report, there may be some benefit from using more than one NLP, as long as there is a significant difference between the two NLPs. This is true with the Stanford and MINIPAR NLPs.
-
Stanford Parser This parser is designed by the NLP Group at the University of Stanford. The parser is rule-based and uses an A* search parsing algorithm to perform efficient and exact inference on separately scored semantic and syntactic parse trees. The parser uses the Penn Treebank corpus MINIPAR This parser was designed by Dekang Lin at the University of Alberta. MINIPAR is a principal-based parser. It uses principals to constrain X-bar structures
I plan to only use one of these NLPs im my initial implementation and make using two NLPs an adaptation. I will probably use the Stanford Parser as my initial NLP, as its output is much easier to understand, therefore putting it into a data stucture and working with that data structure should be simpler.